Natural hybridisation between Aphis grossulariae and Aphis Schneideři: morphological evidence (Sternorrhyncha: Aphididae)

Morphometric analysis of 176 natural samples of A. grossulariae Kaltenbach, 1843 and A. Schneideři (Borner, 1940) was performed, using 308 alate and 750 apterous viviparous females from 25 countries altogether. Morphologically intermediate speci­ mens of presumably hybrid origin were noticed in 63 (35.79%) samples, comprising 12.67% of all apterous and 4.87% of all alate viviparae studied. 31 sample originating from 11 countries had 50% or more intermediate specimens of one or both morphs. “Rich” samples (having 4 or more specimens of the same morph) with the prevailing numbers of hybrid morphotypes were from the Nether­ lands, Russia (Moscow and Stavropol regions), Moldova, Turkey (Ankara) and Tajikistan (Dushanbe). Present data are discussed in the context of possible natural hybridisation between A. grossulariae and A. schneideri (see also Rakauskas, 1999a, 1999b). Canonical discrimination functions are being advocated as more powerful tools for separating between the two species when com­ pared with single morphological characters or ratios commonly used in the keys.


INTRODUCTION
The possibility of hybridising and producing viable and fertile progeny is an important feature of biparental spe cies, being emphasized by the reproductive species defi nitions, including the biological (e.g.Mayr, 1982;Dobzhansky, 1970) and recognition species concepts (Paterson, 1993).Various isolating mechanisms usually prevent natural hybridisation between "good" species in nature.Therefore, hybridisation studies might supply important information on the taxonomic status of the forms involved in a complex (Müller, 1985;Shaposhnikov, 1987;Guldemond, 1990;etc.).
Palaearctic species of the genus Aphis L. inhabiting currants [A. grossulariae Kaltenbach, 1843, A. triglochinis Theobald, 1926, and A. schneideri (Börner, 1940)] have been reported as capable of interspecific hybridisation under experimental conditions (Rakauskas, 1999a(Rakauskas, , 1999b(Rakauskas, , 2000;;Turcinaviciené, 2000).Experimental crosses normally reproduced by means of parthenogene sis, and fertile bisexual generations appeared in some cases.When pondering whether currant inhabiting Aphis species are capable also of natural hybridisation, mor phometric methods can provide certain information.Namely, 26% of experimental A. grossulariae x A. schneideri hybrid clones appeared to be morphologically intermediate between the parental species (Rakauskas, 1999a).Such intermediate hybrid morphotypes can be easily detected in nature thus supporting the idea of natural hybridisation ofboth species.
The aim of this work was to perform morphometric analysis of Aphis grossulariae and A. schneideri samples from various collections seeking for the natural hybrid morphotypes having intermediate morphological charac ters.

MATERIAL AND METHODS
Morphometric data have been extracted from specimens labelled as A. schneideri and A. grossulariae in various public and private collections in 1987-2001.Currant specimens in the collection of Natural History Museum (London) that have been labelled by D. Hille Ris Lambers as new species Aphis szelegiewiczi and Aphis subepilobii respectively (unpublished nomina nuda), have been also analysed.Only currant morphs of A. grossulariae [collected on currants and gooseberries (Ribes spp.)] were used.176 samples comprising 308 alate and 750 apterous viviparous females from 25 countries have been studied altogether (Table 1, Fig. 1).The whole list of analysed samples is available from the author on the request.
Typical morphological features of both species were taken from clonal specimens.Clones that were holocyclic faculta tively heteroecious between Ribes spp.and Onagraceae herbs (13 clones, 105 apterae and 100 alatae altogether) were taken as typical A. grossulariae.Holocyclic monoecious on Ribes spp.clones were taken as typical A. schneideri.(17 clones, 113 apterae and 102 alatae altogether).These clones were originally started from Polish and Lithuanian material.The entire list of typical clones and their detailed morphometric data are pub lished (Rakauskas, 1993(Rakauskas, , 1998)).These data were used for obtaining key characters and canonical discrimination function values typical for A. schneideri and A. grossulariae.
Two methods were used for the morphological identification of museum specimens.First, the identification was attempted using key characters (Rakauskas, 1998).For example, apterous viviparous females having ratio siphon length: longest hair on third antennal segment length exceeding 7 were identified as A. grossulariae.Specimens having the same ratio smaller than 6.6 were taken as A. schneideri.Ratio values from 6.61 to 6.99 were recognised as intermediate.
Second method was canonical variates analysis, a method that has proved very useful in distinguishing closely related aphid species (e.g., Blackman, 1987Blackman, , 1992)).Morphometric data of the above mentioned typical A. schneideri and A. grossulariae clones (30 morphological characters, the entire list published in Rakauskas, 1998) were used for calculating canonical discrimi-Fig.1. Sampling places of A. schneideri, A. grossulariae (both from Ribes spp.and summer hosts) and hybrid specimens.Locali ties, where experimental hybrid clones were received (Rakauskas, 1999a(Rakauskas, , 1999b)), are marked with the asterisks.nation functions (CDF) for every morph.Variables to be used in the CDF were selected on the basis of their discriminatory power: those having the smallest partial Wilks' Lambda were taken when calculating CDF for every morph (for details see StatSoft, 2000, Vol.III, Chapter 2).The list of variables used when calculating CDF for every morph (with the respective constants and variable coefficients) is presented in Table 2.The obtained CDF values were subsequently counted for every museal specimen of respective morph.Scatterplot of the CDF individual values of apterous viviparous females from 12 sam ples representing eight countries is presented in Fig. 2, respec tive box and whisker plot of the same samples -in Fig. 3. Information on the morphological features of apterous and alate viviparous females was summarized, an example being pre sented in Table 3.
All calculations were done using the STATSOFT statistical package STATISTICA for WINDOWS 5.5 (StatSoft, 2000).

RESULTS AND DISCUSSION
63 samples out of 176 (35.79%) had one or more mor phologically intermediate specimens, 95 apterous vivipa rous females (12.67% of all analysed apterae) and 15 alate viviparous females (4.87%) among them.These samples originated from 15 countries (Table 1, Fig. 1).31 sample had 50% or more intermediate specimens of one or both morphs (Table 3).These were from 11 countries, Russian Far East, Subcaucasus and Moscow regions among them.Relatively numerous samples (with 4 or more specimens of the same morph) having prevailing numbers of hybrid morphotypes were only six (Table 3).Table 3.Samples having 50% and more specimens with intermediate morphological characters of one or both morphs when esti mated by means of the common key characters (Rakauskas, 1998) or the CDF (wider explained in Material and methods).Rich samples with the prevailing hybrid morphotypes are in bold.

Label data
Abbr morphologically similar to one or another parental species.Cases, when specimens of two species are pre sent in the same sample, can be also explained by means of hybridogenesis: morphological splitting inside the hybrid clones has been reported for experimental crosses A. grossulariae x A. schneideri (Rakauskas, 1999a) and A. grossulariae x A. triglochinis (Turcinavicienè, 2000).
In present study there appeared 11 samples (6.21% of all studied) having specimens of two species.This figure, together with the above mentioned 63 samples (35.59%) having intermediate morphotypes supports the idea of the natural hybridisation between A. schneideri and A. gros sulariae.Nevertheless, further studies are needed to con firm this: DNA analysis of pure and hybrid clones (e.g.microsatellites and mitochondrial DNA techniques, see Hales et al., 1997;Sunnucks et al., 1997) would help to evaluate the degree of introgression between natural populations of these species.Natural isolating mecha nisms such as sex pheromone specificity, circadian rhythm of sex pheromone release, and other aspects of possible natural specific mate recognition system should be also studied (Guldemond et al., 1994;Guldemond & Dixon, 1994;Thieme & Dixon, 1996).Morphological features of different morphs did not coincide within the same sample.For example, 6 out of 10 apterae in Moldova (Karmanovo, No. 1943, "m l") sample appeared to be morphologically intermediate, whilst all 8 alatae of this sample had morphology of A. schneideri (when estimated by CDF).Similar situation was also in samples ko1, li, sw, ta1 (Table 3).The only opposite example is a sample from Turkey (Ankara, 3834, "tu"), in which 3 (out of 4) apterae and 2 (of 4) alatae had intermediate morphology when evaluated by means of CDF.Noticeable, that evaluation results were  3).
different when using key characters and CDF.In the above mentioned sample from Ankara the key characters showed all specimens being "good" species, whilst 5 (of 8) specimens appeared to be intermediates when esti mated by means of CDF.Evaluation results when using different methods coincided only in 4 samples out of 31 presented in Table 3.This is due to the reduced discrimi natory power of the key character when compared with CDF.The "gap" between A. grossulariae and A. schneideri when estimated by the CDF comprises 5 units (from -2 till + 3, Table 2) when compared with 0.4 (6.6 till 7) that is supported by the key ratio (Rakauskas, 1998).Respective gaps for the alate viviparous females are even more narrow: 2.2 when using CDF and 0 for the key ratio.That is, alatae of A. grossulariae have the ratio siphon length/longest hair on ant.segm.III length more than 5.5, whilst those of A. schneideri less than 5.5 (Rakauskas, 1998).It can be concluded here that the CDF is more reliable for the identification of the currant inhab iting Aphis species when compared with the common key characters.This has been already shown for the identifi cation of the Dysaphis chaerophyllinabrachycyclica complex (Stekolshchikov & Lobanov, 1990) and Myzus persicae group (Blackman, 1987).
Fig. 2. Scatterplot of the canonical discrimination function individual values plotted against the body length of the apterous viviparous females (sample abbreviations as in Table 3) showing the distribution of respective values in A. grossulariae (N = 105) and A. schneideri (N = 113).

Fig. 3 .
Fig. 3. Box and whisker plot of canonical discrimination function values for the apterous viviparous females of A. grossulariae (N = 105) and A. schneideri (N = 113) and museum samples (sample abbreviations as in Table3).

Fig. 4 .
Fig. 4. Scatterplot of the canonical discrimination function individual values plotted against the body length of apterous viviparous females of the new species of D. Hille Ris Lambers and A. varians Hille Ris Lambers nec Patch showing the distribution of respec tive values in A. grossulariae (N = 105) and A. schneideri (N = 113).

Table 1 .
Aphid material used in this study showing the num bers of analysed samples, alate and apterous viviparous females (% of intermediate specimens or samples having intermediate specimens).

Table 2 .
Canonical discrimination functions (CDF) constants and coefficients for the discrimination between apterous and alate viviparous females of A. schneideri and A. grossulariae.